Bonjour Sean,

Le mercredi 15 septembre 2004 à 2:29:31, vous écriviez :

>> The more complex some software piece is or any engineering design is,
>> the more weak it is, very generally speaking.

> The FB engine is very complex; that does not mean that it is weak.
> I don't think that "weak" is the word you meant to use, in the context:
> weak == poor quality.

>> It is then fair to question wether the new NBackup system meet real-life
>> expectations in terms of robustness.

> I agree. It is completely fair to ask these questions. But *without*
> performing a real-life review of any functionality, it is unfair to
> ask questions which suggested that the function is poorly designed.
> Since you said "weak" this is what I took from your post and reacted
> to.

Then my english language knowledge played tricks on me and you. I
didn't intended 'poor quality' by selecting the word 'weak'. I just
checked my dictionnary and indeed the french translations for weak say
'faible', 'vulnérable' and even 'débile', which is on the edge of
being impolite. This is a parsec away, and more, from what I wanted to
write. Let's say 'weak' was, in my mind, a kind of opposite to
'robust'.

Complex assemblies can be robust: when they have no design flaws and
after implementation bugs have been wiped out. Until then, I tend to
consider that the simplest of two approaches has potential to be more
robust. I have no reasons to suspect design flaws in what I know from
NBackup for now. But I can legitimely suspect, even programmed by
Nicholay or whoever else, that the implementation complexity might
reveal bugs in NBackup, or worse, in the engine behaviour. Newcomers
will question that. They will want to know for how long this backup
technology has been in use, and how much confidence other people have
put in it by using it for real and having their business data safety
rely on it.

>> I suppose you have at Broadview some success
>> stories to tell about NBackup. I have none yet because I couldn't yet
>> decide a customer to agree to run FB 2.0 codebase in production.

> We aren't running in production, yet. We have it deployed on our main
> development server, on which we run 10-12 client databases.

> It is our plan to deploy a private v2.0 build to a selected client site
> in the next 1-2 months, and then deploy more broadly. (The build will
> not have any private functions, it will just be built from a very, very
> well tested copy of the HEAD)

Good news. I also intend to switch developers here on head branch soon
(mid-october is our target). That will allow us to play with new
features and Nbackup on copies of real users databases.

>> For NBackup to be a real bonus to Firebird, it will need more than
>> documentation. It will need useage case scenarios...

> I could not agree more strongly.

I then think we are perfectly in-sync with this, I just slipped on a
word whose exact accepted meaning in english wasn't what I thought it
was.

Thanks !

--
Cordialement,
Olivier Mascia